Loan loss reserves: The pandemic in five charts

One of the more notable stories of the big US bank quarterly earnings seasons this year has been loan loss allowances. The reserves that the banks had ramped up in the early days of the pandemic are starting to be released.

Taken in aggregate, loan loss allowances at Bank of America, Citigroup, Goldman Sachs, JPMorgan, Morgan Stanley and Wells Fargo soared from $57.5 billion on January 1, 2020, to about $101 billion six months later, a rise of 75%.

And there they peaked, sitting at the same level through the third quarter of the year.

The numbers were dominated by the money centre banks, and JPMorgan more than most. Fully one-third of the increase – some $14 billion – was built at JPMorgan alone. The next biggest contributors were Wells Fargo, at 24%, and Citi, at 21%.

Since then, starting in the fourth quarter of 2020 but gathering pace in the first half of 2021, much has been made of the fact that banks are releasing those reserves, something undoubtedly made more possible by continued central bank stimulus and government support schemes that are staving off the worst of the pandemic-related pain to businesses and households.

But here’s the first thing: loan loss allowances haven’t fallen by as much as you might think. Yes, they are down by 30% from their peak, but they still stand 25% above where they were on January 1, 2020. On that basis alone, it doesn’t look like banks are close to declaring pandemic risk over.

A couple of other points are worth noting. First, Wells Fargo’s allowances have not fallen anything like as much as at the other big banks. It’s easy to forget now that at the start of the pandemic there was another crisis already going on in energy markets, something to which Wells is quite exposed.

In its first quarter 2020 earnings, it was its oil and gas sector exposure that Wells called out as a driver of its allowances as well as Covid-19. And that was after Wells Fargo had been one rare example of a big bank able to record a decrease (of $1.3 billion) in its allowances as a result of adopting the current expected credit losses (CECL) standard for loan losses on January 1, 2020.

Once the pandemic fully took hold, Wells’ allowances peaked at the end of the third quarter of 2020 at 2.3 times what they had been on January 1, at nearly $19.5 billion.

Of the other outliers, Morgan Stanley’s absolute allowances are tiny, making the pre- and post-pandemic comparison less meaningful. But Goldman’s are interesting, standing at $3.27 billion, 50% above their pre-pandemic level, having peaked at 80% above. Here the rise reflects the fact that the pandemic came just as Goldman was pushing into new areas of consumer banking, with its associated risks.

Which brings us to our next charts. There are, of course, two functions to loan loss allowances: assessment of risk is one; the other is absolute levels of lending. So, what’s happening there?

Aggregate volume of retained loans has fallen slightly, at $3.67 trillion, to stand at about 97% of what it was pre-pandemic. It rose in the first quarter of 2020 – by about $250 billion – as clients sought the security of cash from their facilities. But of the big lenders, only JPMorgan has increased its overall lending since the start of 2020, and then only by 2%.

Where lending is up a lot is at the relative minnows of Goldman Sachs and Morgan Stanley. At Goldman, loans are up by 34%, to $121 billion; at Morgan Stanley, they are up 31%, to $155 billion.

What does all this tell us about risk? We have lending down a bit, but allowances that are still elevated. That adds up to more caution than before – about 30% more, in fact. The aggregate loan loss allowance to retained loans ratio was 1.5% at the start of the pandemic. Now, after peaking at 2.7%, it’s standing at just under 2%.

Indexed to a 100 base on January 1, 2020, aggregate loan loss allowance ratios now stand at 129.

Even that number hides other stories. The aggregate level is being dragged up by a much higher allowance ratio at Wells Fargo, whose indexed ratio stands at 205. And what stands out even more is that this is even after Wells has reduced its aggregate lending by more than any other of the big banks, to just 89% of where it was pre-pandemic.

Morgan Stanley is the other outlier, with an indexed ratio of 147, the only other firm to be above the aggregate level, although that comes after the lending increase already noted.

The CECL effect

The other factor thrown into the mix is CECL. The fourth quarter 2019 levels used in all the charts above reflect that January 1 adoption, which mostly saw banks increase their allowances from what they had previously reported. And the more rigorous approach helped to drive the increases that would come in subsequent quarters as a result of the pandemic.

The big question is: what is the end game? All things being equal, as conditions normalize, should the day-one CECL allowance ratios be the mean to which banks revert over time?

Some bankers think that might not be right. One US bank CEO tells me: “Is day-one CECL the proper stopping point now? It’s possible you cross that threshold in future.”

The reason is the models that power CECL itself. While the new standards were hailed as being more stringent for the fact that they looked through to the expected losses over the lifetime of a loan, they were also criticised for adding more pro-cyclicality, meaning that allowances would inevitably be more volatile than before.

What’s critical here is that when CECL was first adopted, there hadn’t been a US recession for more than 10 years, not since the global financial crisis. But every quarter banks reassess the forward economic assumptions that they plug into the model. While this made it important to factor a possible recession into day-one CECL values, that’s less compelling now that it has happened. Unless the risk profile of your lending has shifted, it’s logical that allowances – and ratios – might drop to below pre-pandemic levels.

Net charge-off data seem to support this. For our six banks, net charge-offs peaked at $6.2 billion for the second quarter of 2020, up from $5.3 billion in the last quarter of 2019. By the second quarter of 2021 they had fallen to just $3 billion.

Is this just more evidence of pain delayed rather than avoided? Bankers don’t seem to think so. Some say much of the corporate pain has already been shaken out: some companies have failed; those that haven’t have worked on their cost bases; others have thrived.

For households, the calculation may be a little more nuanced. Loan forbearance, income support schemes and lower spending have left many not much worse off or else with considerable cash balances. Some of that does surely reflect pain that is delayed, but not all.

But even that is not the whole story. Bankers are now grappling with the prospect of consumer loan losses that will increase not because individuals’ pandemic-related financial difficulties have been postponed but more because that will be a natural result of consumers beginning to spend – and overspend – again.

It might be tempting to see economic recoveries as easier to risk-manage than crises, but it doesn’t look that way yet.